New Memory Causing Filesystem Failure

gryffin · 2024-03-25 10:33:57

I recently bought two sticks of Corsair Vengeance DDR4 8GB RAM, identical make and model to the 2x8 RAM I have been using for years. When I insert the new memory, after powering on my computer, I am getting errors with one or both of my SSDs.

My filesystem uses BTRFS and is hosted on two M.2 SSDs, nvme0n1 and nvme1n1.

Immediately after powering on my system, I get btrfs warnings and errors concerning nvme1n1:

Mar 24 21:40:52 geordi kernel: BTRFS warning (device nvme1n1p2): csum failed root 256 ino 5624634 off 32788480 csum 0x318bfd7a expected csum 0x7b3e82c7 mirror 1
Mar 24 21:40:52 geordi kernel: BTRFS error (device nvme1n1p2): bdev /dev/nvme0n1p1 errs: wr 0, rd 0, flush 0, corrupt 24, gen 0

Mar 25 02:24:15 geordi kernel: BTRFS warning (device nvme1n1p2): csum failed root 256 ino 2935302 off 2359296 csum 0x0c6ce243 expected csum 0x8eeb7e90 mirror 1
Mar 25 02:24:15 geordi kernel: BTRFS error (device nvme1n1p2): bdev /dev/nvme0n1p1 errs: wr 0, rd 0, flush 0, corrupt 25, gen 0

If I shut down the system and remove the two new RAM sticks, these errors do not appear. When I shut the system down and re-insert the new RAM sticks, the errors re-appear.

Normally, the result of lsblk is as follows:

[griffin@geordi ~]$ lsblk -f
NAME        FSTYPE FSVER LABEL  UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
zram0                                                                               [SWAP]
nvme1n1                                                                             
├─nvme1n1p1 vfat   FAT32        F5EC-C26C                             448.6M    12% /boot
└─nvme1n1p2 btrfs        geordi 1fee3a92-b9e2-4ef9-83fd-b1ffa9dd0b3e  661.5G    64% /var/log
                                                                                    /var/cache/pacman/pkg
                                                                                    /home
                                                                                    /.snapshots
                                                                                    /
nvme0n1                                                                             
└─nvme0n1p1 btrfs        geordi 1fee3a92-b9e2-4ef9-83fd-b1ffa9dd0b3e

When the two extra ram sticks are inserted, lsblk only detects the nvme1n1 device, and the nvme1n1p2 btrfs partition is labeled nvme1n1p1, with nvme1n1p2 vfat partition, and the nvme0n1 device not shown.

When the ram is inserted, btrfs check health /dev/nvme1n1p1 fails:

1 root@archiso ~ # btrfs check /dev/nvme1n1p1
Opening filesystem to check...
warning, device 1 is missing 
warning, device 1 is missing
bad tree block 2200120147968, bytenr mismatch, want=2200120147968, have=0
ERROR: cannot read chunk root
ERROR: cannot open file system

When the ram is removed, btrfs check returns no errors:

1 root@archiso ~ # btrfs check /dev/nvme0n1p1
Opening filesystem to check...
Checking filesystem on /dev/nvme0n1p1
UUID: 1fee3a92-b9e2-4ef9-83fd-b1ffa9dd0b3e
[1/7] checking root items
[2/7] checking extents
[3/7] checking free space tree
[4/7] checking fs roots
[5/7] checking only csums items (without verifying data) 
[6/7] checking root refs
[7/7] checking quota groups skipped (not enabled on this FS)
found 1270483660800 bytes used, no error found
total csum bytes: 1235402116
total tree bytes: 4574052352
total fs tree bytes: 2896904192
total extent tree bytes: 232980480
btree space waste bytes: 802475244
file data blocks allocated: 5620576358400
 referenced 1294018232320
btrfs check /dev/nvme0n1p1 14.91s user 4.09s system 41% cpu 45.725 total

What is happening here? Is this a software or hardware issue? Is there something I'm doing wrong that's caused this?

jl2 · 2024-03-25 11:20:56

Do you boot with 2 sticks or 4 sticks?
if yes, have you tested only the 2 new sticks, or can you pin down 1 stick (try booting only 1)?

This somehow sounds like faulty memory, and somehow doesn't.
Can you also post lsmem?

Last edited by jl2 (2024-03-25 11:21:42)

gryffin · 2024-03-25 11:39:03

jl2 wrote:

Do you boot with 2 sticks or 4 sticks?
if yes, have you tested only the 2 new sticks, or can you pin down 1 stick (try booting only 1)?
This somehow sounds like faulty memory, and somehow doesn't.
Can you also post lsmem?

I have tried booting in two ways: First, with only the original 2 sticks of RAM (the "control" state). Second, with 4 sticks of ram, the original plus 2 new sticks (the "failing" state)
I'm not sure what you mean by pin down 1 stick, but I can try booting it with only one of the new sticks and none or both of the old ones.

[griffin@geordi ~]$ lsmem
RANGE                                  SIZE  STATE REMOVABLE  BLOCK
0x0000000000000000-0x00000000dfffffff  3.5G online       yes   0-27
0x0000000100000000-0x000000041fffffff 12.5G online       yes 32-131

Memory block size:       128M
Total online memory:      16G
Total offline memory:      0B

This is lsmem in the control state, without any crashing behavior. I will try more combinations, memtest too, tomorrow.

jl2 · 2024-03-25 19:59:29

I meant if the problem also emerges if you only use the 2 new sticks, and if it emerges if you try the new sticks one at a time.
And the lsmem of all 4 sticks?

Last edited by jl2 (2024-03-25 20:00:06)

seth · 2024-03-25 21:14:33

memtest86+ - overnight/16h+ (at least or until you hit errors)

Generally just try to use the slowest clocks/timings you can configure for the RAM and post the output of "sudo dmidecode" - unless you've same vendor,model and even batch, you cannot run mixed DIMMs at maximum performance.
Also consult your board manual, it'll tell you where to put them (the order becomes relevant) and often they're color-coded (so typically put the same kind of RAM into the same colored slots - your boar manual is authorative here, though)

gryffin · 2024-03-25 23:59:14

I ran memtest86 this morning with the new RAM installed and saw a bunch of errors. I ran it with only my old RAM and there were no errors. I feel comfortable assuming the problem is faulty memory. I'll return to this thread if a replacement continues to cause errors. For anybody who's interested, I do not overclock my PC. My motherboard is an ASRock B450 Pro4. My original RAM's part number is CMK16GX4M2A2400C16, and the new RAM has part number CMK16GX4M2A2400C16. Thanks to everyone in this thread for your advice and suggestions.

Last edited by gryffin (2024-03-26 02:53:30)

redshoe · 2024-03-26 02:23:25

100% memory fault. I had similar experience.

Arch Linux

#1 2024-03-25 10:33:57

New Memory Causing Filesystem Failure

#2 2024-03-25 11:20:56

Re: New Memory Causing Filesystem Failure

#3 2024-03-25 11:39:03

Re: New Memory Causing Filesystem Failure

#4 2024-03-25 19:59:29

Re: New Memory Causing Filesystem Failure

#5 2024-03-25 21:14:33

Re: New Memory Causing Filesystem Failure

#6 2024-03-25 23:59:14

Re: New Memory Causing Filesystem Failure

#7 2024-03-26 02:23:25

Re: New Memory Causing Filesystem Failure

Board footer